Threats to Energy and Water Security for a Colorado River ... · 11/13/2018 · Case Study of a Provisioning Basin for Water and Energy: The San Juan Basin • A key source of water

Threats to Energy and Water

Security for a Colorado River Basin

Provisioning Watershed

• Moderator: Dagmar Llewellyn, Bureau of Reclamation,

Albuquerque

• Speakers:

– Katrina Bennett: Los Alamos National Laboratory, NM

– Susan Behery: Bureau of Reclamation, Durango, CO

– Lucas Barrett: Bureau of Reclamation, Albuquerque, NM

– Ariel Miara: National Renewable Energy Laboratory, Golden,

CO

– Sheh-Yu Huang: Lehigh University, Pennsylvania

Case Study of a Provisioning Basin for

Water and Energy: The San Juan Basin• A key source of water to the Colorado and Rio Grande

basins, for:

– Agriculture /food security,

– Tribes (Navajo Nation, as well as downstream tribes in both

basins)

– municipalities

– recreation, and

– ecosystems.

Navajo Organics

Case Study of a Provisioning Basin for

Water and Energy: The San Juan Basin

• A key source of water-dependent energy generation for

distribution throughout the southwest.

– Oil and gas extraction, and

– Energy generation (thermal electric power and hydropower

generation)

Messages from this suite of

presentations….

• Basins are connected through transfers of both

water and energy.

• Water supplies affect energy generation, and energy

is needed to transport water.

• The San Juan Basin (a sub-basin of the Upper

Colorado) serves as a provisioning basin for both

water and energy – changes in provisioning basins

can reverberate throughout the region.

• Interdisciplinary teams, often from multiple agencies

and institutions, are needed to evaluate these

complex interactions.

Operated by Los Alamos National Security, LLC for the U.S. Department of Energy's NNSA

Katrina E. Bennett

Earth and Environmental Sciences

Los Alamos National Lab, Los Alamos, NM

Upper Colorado River Basin Forum

Grand Junction, Colorado

Nov 7th, 2018

LA-UR-

Upper Colorado River Basin ForumGrand Junction, Colorado Nov 7th, 2018

Susan Behery, P.E.

Reclamation

Western Colorado Area Office

Durango, Colorado

Implications for Water Supply in

the Rio Grande Basin

Lucas Barrett – Bureau of Reclamation,

Albuquerque Area Office7

Linking Climate, Water Resources And Electricity Generation:A Case Study Of The San Juan River Basin And Its

Transboundary Impacts On The Western Interconnection

Ariel Miara, National Renewable Energy Laboratory

2018 Upper Colorado River Basin Water Forum: Bridging Science, Policy & Practice

Grand Junction, CO

November 7, 2018

Sandia National Laboratories, Los Alamos National Laboratory, Lehigh University, University of Colorado-Boulder: Center for Advanced Decision Support for Water and Environmental

Systems, Pacific Northwest National Laboratory, City University of New York, University of Southern California, Bureau of Reclamation

Complex Adaptive Water Systems (CAWS) Research Group

Department of Civil and Environmental Engineering

Shih-Yu Huang (Lehigh),

Y. C. Ethan Yang (Lehigh),

Vincent C. Tidwell (Sandia)

11/07/2018

Development of a

coupled RiverWare-

agent-based model

for the San Juan

River Basin

Katrina Bennett

Los Alamos National Laboratories,

New Mexico

Operated by Los Alamos National Security, LLC for the U.S. Department of Energy's NNSA

Katrina E. Bennett

Earth and Environmental Sciences

Los Alamos National Lab, Los Alamos, NM

Upper Colorado River Basin Forum

Grand Junction, Colorado

Nov 7th, 2018

LA-UR-28766

Drought

Insects

Background

11/13/2018 | 12Los Alamos National Laboratory

Large forest fires around Los

Alamos in 2002 & 2011

Drought

Insect mortality

Large floods in 2011 & 2013

WildfireMo

rta

lity D

rive

n

Dis

turb

ance

Re

sultan

t

Hydro

logic

al

Impa

cts

Study Area – Upper Colorado River sub-basin

11/13/2018 | 13

Los Alamos National Laboratory

San Juan covers 22% Upper CRB

15% of basin water to Lees Ferry

Methods – Climate Scenarios


CMIP5 earth system models

(ESMs) with dynamic vegetation

(RCP 8.5)

HadGEM2-ES

MPI-ESM-LR

MIROC-ESM

IPSL-CM5B

IPSL-CM5A

CanESM2

Downscaled using MACA -

Multivariate Adaptive Constructed

Analogs (Abatzoglou and Brown,

2011)

Annual Differences (2080s - 1980s)

Temperature (°C)

Pre

cip

itation (

%)

GCMs/ESMs

CAN

MIR

HAD

IPA

IPB

MPI

0

VIC Version 4.2x

MODIS Veg;

Livneh et al.

2015

Methods – Hydrological Modeling


VIC 1/16th of a degree grid scale, 1-hr time step

MODIS based vegetation parameterization, time

series of each grid cell for albedo, vegetation

spacing and LAI

Clumped vegetation formulation, appropriate

for semi-arid Southwest

Bohn and Vivoni, 2016, WRR. Bennett et al. 2018. WRR. Global sensitivity of simulated water balance indicators

under future climate change in the Colorado Basin.

Methods – Model Calibration


• Calibrated model using a multi-objective calibration approach (Yapo et

al. 1998)– Three objectives: Nash-Sutcliffe (peaks), log Nash-Sutcliffe (low flow), mean annual volume bias

– Calibrated model for baseflow indicators, soil depths, and new snow albedo

Basin ID and NameUSGS

Gaugea nRMSEb

Nash

Sutcliffe

Efficiency

Gunnison River below Blue Mesa Dam, CO 09124700 0.71 0.55

Yampa River near Maybell, CO 09251000 0.50 0.87

Green River below Fontenelle Reservoir, WY 09211200 0.66 0.64

Colorado River at Glenwood Springs, CO 09072500 0.27 0.94

Gunnison River near Grand Junction, CO 09152500 0.42 0.80

White River near Watson, UT 09306500 0.34 0.82

Dolores River near Cisco, UT 09180000 0.44 0.83

Green River near Green River, WY 09217000 0.35 0.89

Green River near Greendale, UT 09234500 0.37 0.87

Colorado River near Cisco, UT 09180500 0.31 0.91

Green River at Green River, UT 09315000 0.29 0.92

San Juan River near Bluff, UT 09379500 0.45 0.76

Colorado River at Lees Ferry, CO 09380000 0.22 0.95a Modeled streamflow calibrated using naturalized streamflow data at specified locationb RMSE normalized to observation mean

Solander, Bennett, et al. 2018. In review. Estimating the vulnerability of mountain hydrology to climate change using historical data.

Methods – Vegetation Scenarios


2000-2005

Forested

(MODIS)

McDowell et al. Nature Climate Change. 2015.

Convergent predictions of massive conifer mortality due to

chronic temperature rise.

70

80

90

20 30 40 50 60 70 80 90

Decades

HadG

EM

2−

ES

Fore

sts

(%

)

20

40

60

80

dec

DecadesHa

dG

EM

2-E

S F

ore

st (%

)

CMIP5 vegetation

dynamic models

Empirical /physically

based models for the

US Southwest

70

80

90

Bennett et al., 2018. HESS. Climate-driven

disturbances in the San Juan River sub-basin of the

Colorado River

Susan Behery

Bureau of Reclamation, Durango,

Colorado

Upper Colorado River Basin ForumGrand Junction, Colorado Nov 7th, 2018

Susan Behery, P.E.

Reclamation

Western Colorado Area Office

Durango, Colorado

VIC and RiverWare Natural System

EngineeredSystem

RiverWare Model

VIC Model

Diverse and Competing Water Use

Lucas Barrett

Bureau of Reclamation,

Albuquerque, New Mexico

Implications for Water Supply in

the Rio Grande Basin

Lucas Barrett – Bureau of Reclamation,

Albuquerque Area Office25

Inter-Basin Transfer to the Rio

Grande • Reclamation’s San Juan-Chama Project transfers a

portion of New Mexico’s allocation of the Upper Colorado

River from three tributaries of the San Juan to the water-

short Rio Grande Basin, where it forms a critical portion

of the supply.

• Hundreds of thousands of water users in the Rio Grande

basin in New Mexico are reliant on this transbasin water

supply. In the summer of 2018, all native water supplies

in the Rio Grande were depleted, and the Middle Rio

Grande was entirely reliant on stored supplies from the

San Juan-Chama Project.

26

San Juan-Chama Project Overview

• A Participating Project of the

Colorado River Storage Project,

authorized by amendment to

the Colorado River Storage Act

of 1956.

• Primary project purpose is

water supply to the Middle Rio

Grande Valley for irrigation,

municipal, domestic, and

industrial uses

• Also authorized to provide

incidental recreation and fish

and wildlife benefits27

Oso Diversion, Chromo, CO

Azotea

Tunnel

Outfall,

Willow

Creek,

NM

San Juan-Chama Project Map

28

Heron Dam: Equalizes Year-to-Year Supply to the Rio

Grande Basin

29

30* Water used exclusively for irrigation.

Who Relies on

Water Supply

from

Reclamation’s

San Juan-

Chama

Project?

Katrina Bennett

Los Alamos National Laboratories,

New Mexico

Results – Hydrological Flows


• Historical flows peak in late May,

early Jun

• “Full Use” and “Future Climate”

– -> streamflow declines of ~33%

– Greatest impacts spring, summer and

early fall

• “Future Climate” -> early spring

melt and peak flow, fall flows

higher, annual declines of ~14%

• “Future Climate and full use, 2018

water year shown for comparison

Bennett, Tidwell, Llewellyn, Behery, Barrett, et al. 2018, in prep. Threats to a Colorado River

Provisioning Basin from Coupled Future Climate and Societal Changes.

Solander, Bennett, et al. 2018. JHM. Interactions between Climate Change and Complex

Topography Drive Observed Streamflow Changes in the Colorado River Basin.

Results – Navajo Storages and Flow to Powell


• Navajo flows do not cross over

shortage sharing cutoff

• Low points that are even lower

after the 2030s, particularly

under climate and full use

• Flows to Powell fall below

minimum flow requirements

(500 cfs), declining flows with

some extremes after 2005s

• Range across models depicts

uncertainty in ESM projections,

reduced under operational

constraints

Bennett, Tidwell, Llewellyn, Behery, Barrett, et al. 2018, in prep.

Threats to a Colorado River Provisioning Basin from Coupled

Future Climate and Societal Changes.Minimum flow requirements

Results – Shortage Sharing


a)

Full UseW

ate

r S

hort

ed

San J

uan (

%)

25

50

75

HadGEM2−ES

MIROC−ESM

MPI−ESM

CanESM2

IPSLA

IPSLB

Climate

b)

Climate and Full Use

c)

d)

25

50

75

Wate

r S

ho

rte

d

San J

ua

n−

Cha

ma (

%)

e) f)

g)

−2

02

46

Flo

w T

arg

et M

et (%

)

1970−1999

h)

2070−2099

i)

2070−2099

Results - Uncertainties


−1

00

−50

050

short SJ short SJC navajo srelease powell

Diffe

ren

ce

s (

%)

to 2

080

s

Future Full Use minus Historical

Future Climate and Full Use minus Future Full Use

Future Climate minus Historical

Short S. Short S. Navajo Spring Powell

San Juan Chama Release

Diffe

rence (

2080s,

%)

Ariel Miara

National Renewable Energy Labs,

Golden, Colorado

Linking Climate, Water Resources And Electricity Generation:A Case Study Of The San Juan River Basin And Its

Transboundary Impacts On The Western Interconnection

Ariel Miara, National Renewable Energy Laboratory

2018 Upper Colorado River Basin Water Forum: Bridging Science, Policy & Practice

Grand Junction, CO

November 7, 2018

Sandia National Laboratories, Los Alamos National Laboratory, Lehigh University, University of Colorado-Boulder: Center for Advanced Decision Support for Water and Environmental

Systems, Pacific Northwest National Laboratory, City University of New York, University of Southern California, Bureau of Reclamation

NATIONAL RENEWABLE ENERGY LABORATORY 38

Electric generators are affected by water-climate constraints

Power plants shut down or curtail output due to high water temperatures or lack of water


Output comparing hydropower dispatch from an energy model (PLEXOS) and a water model (RiverWare)

Hydropower provides flexibility benefits to the grid that are not always captured by many grid models

Water model Grid model

Ibanez, Eduardo, Timothy Magee, Mitch Clement, Gregory Brinkman, Michael Milligan, and Edith Zagona. 2014. “Enhancing Hydropower Modeling in Variable Generation Integration Studies.” Energy 74:518–528

• Energy-Water Coupling Challenges

o Modeling objectives

o Temporal dimensions

o Geographic scales

o Downstream impacts

o Hydro operating rules

o Water storage

o Degree of coupling

Linking energy and water models can lead to new insights: Case study with hydropower flexibility benefits


How do local changes in water deliveries, hydropower, and thermal power plants impact local and regional power generation?

• Realistic physical and legal water constraints on water deliveries to basin power plants

• Impacts to power plant operations represented as inputs into power system models (variable, time & spatial scales)

• Captures local power system dynamics due to localized water changes

• Captures regional power system dynamics due to localized water and energy changes

Approach requires multiple models linked across temporal and spatial scales

Watershed Scale

Runoff

RiverWare

VIC

PLEXOS

Climate Scenarios

Land Use / Land Cover Change

Scenarios

Socioeconomic &

Population

Scenarios

Energy System

Scenarios

Water Shortages

RiverWare

PLEXOS-local

PLEXOS-regional

Creation of new Power Plant Diversion Object in RiverWare


How will San Juan River Basin plants and broader WECC respond to water shortage sharing agreements in the San Juan River Basin?

o 10 scenarios of different climates and water availabilityo Generation within San Juan region and WECC

– San Juan accounts for 3.5% of WECC generation (30/830 TWh/yr)

o Inter-region dependencies

Case study of the San Juan River Basin: What to Expect

Local responses to water shortages by plant

Regional responses

San Juan Basin Schematic


Scenario analysis to evaluate climate impacts, land-use changes, and water demand impacts

Key Question: How do local changes in water deliveries, hydropower, and thermal power plants impact local and regional power generation?

- Scenario drivers:- Climate projections of water availability (C)- Local water demand growth (G)- Vegetation disturbance and land-use change (V)

- Temporal resolution: 1 year modeled at daily resolution- Climate years: 2044, 2064, 2084 (low, medium, high warming

levels)


ScenarioClimate Model Study Year

Water Demand Growth

Vegetation

Four Corners Annual Water Withdrawal

Limit (Million Gallons)

San Juan Annual Water Withdrawal

Limit (Million Gallons)

Base NA 2010 N N NA NA

2044-CHadGEM2-

ES365 2044 N N No Limit No Limit

2044-CGHadGEM2-

ES365 2044 Y N No Limit No Limit2044-CGV

HadGEM2-ES365 2044 Y Y No Limit No Limit

2064-CHadGEM2-

ES365 2064 N N No Limit No Limit

2064-CGHadGEM2-

ES365 2064 Y N 6,715 9,9412064-CGV

HadGEM2-ES365 2064 Y Y 6,423 9,508

2084-CHadGEM2-

ES365 2084 N N 7,161 10,600

2084-CGHadGEM2-

ES365 2084 Y N 2,022 2,9942084-CGV

HadGEM2-ES365 2084 Y Y 1,426 2,111

Scenario analysis to evaluate climate impacts, land-use changes, and water demand impacts


Local generation responses to climate-water changes

Water shortages lead to greater ramping of thermal plants

Under severe droughts, thermal plants are significantly reduced

Some hydro generation is unaffected by severe drought

Ramping capability is reduced under some scenarios

Thermal Plants Hydroelectric


Localized climate-water changes affect local power generation patterns and the impacts cascade to other regions

Baseline Electricity Generation

Changes in electricity generation due to water availability constraints associated with drought and competing users


West-wide generation responses to local climate-water changesD

iffe

rence

in a

nnual genera

tion in W

EC

C b

y f

uel

for

all

sce

na

rio

s, co

mp

are

d t

o b

ase

sce

na

rio

(TW

h)

Meng, M., Macknick, J., Tidwell, V., Zagona, E., Neumann, D., Magee, T., Bennett, K., and R. Middleton. High-resolution integration of water, energy, and climate models to assess electricity grid vulnerabilities to climate change. Applied Energy (in preparation).

WECC-wide, reductions in coal generation in the San Juan River Basin lead to increases in natural gas generation in other regions


It is important to consider:1. Climate impacts on water resources in power system

models to capture hydropower flexibility and thermal plant variability

2. Non-energy water users3. Transboundary impacts of local climate-energy-water

connections

Future Work:1. Expand analysis of local impacts to other watersheds

- Evaluating the Colorado River Basin as a whole with an expanded RiverWare model

2. Incorporate a ‘feed-back loop’ for transboundary impacts- For example, can surrounding regions provide the additional electricity required or are they also constrained?

Closing Thoughts and Future Work

Sheh-Yu Huang

Lehigh University, Pennsylvania

Complex Adaptive Water Systems (CAWS) Research Group

Department of Civil and Environmental Engineering

Shih-Yu Huang (Lehigh),

Y. C. Ethan Yang (Lehigh),

Vincent C. Tidwell (Sandia)

11/07/2018

Development of a

coupled RiverWare-

agent-based model

for the San Juan

River Basin

50/11

Outline

Introduction

Coupled ABM-RiverWare Model

Result

The Way Forward

51/11

Introduction

Science Question: How to better characterize human activities

and integrate those in the energy-water-land modeling at

different spatial scales?

Agent-Based Modeling (ABM)

• Distributed Artificial Intelligence (DAI)

• is driven by its own utility function

• follows behavioral rules

• interacts with other agents and the environment

52/11

Introduction

Why ABM?

Traditional “top-down” approach

• Difficult to properly describe institutional context

• Assume complete information exchange between stakeholders

• Assume no uncertainty in received information

• Adopt fixed sector interaction

Purpose of the study

Develop a coupled RiverWare-agent-based model (ABM) that

can

• characterize human decision-making processes

• quantify the associated decision uncertainty

• assess impacts of water management decisions on the environment

53/11

Coupled ABM-RiverWare Model – ABM

Agents in ABM: farmers within San Juan river basin, how much

water to irrigated is the decision

“Looking backward but acting forward”

An agent’s (farmer’s) decision is affected by

• Internal thinking processes – past experience, new incoming

information, risk perception

• External factors – socioeconomic conditions

54/11

Coupled ABM-RiverWare Model – RiverWare

San Juan River Basin Coupled ABM-RiverWare Framework

55/11

Result – Model Calibration

Muiti-objective Calibration

• Diagnose the model by matching the historical pattern of irrigated area

and streamflow simultaneously

Calibration of Irrigated area Calibration of

Streamflow

56/11

Result – Risk-seeking/averse

How can we use this model to answer the science question?

Scenario: Everyone becomes risk-seeking or risk-averse

Basin-wide Agent level

57/11

Result – Socioeconomic conditions

How do socioeconomic conditions affect decision-making?

Scenario: The cost of expanding irrigation area gradually

increase/decrease through time

Irrigation area changes under different socioeconomic conditions

58/11

The Way Forward

The Watershed scale ABM-RiverWare

• Quantify the impacts of changing climatic and socioeconomic

conditions on water management decisions

• Include reservoir as agent to incorporate reservoir management

• Linking with PLEXOS (through Riverware) to include power plants as

agents

From Watershed scale (ABM-RiverWare) to Regional scale

• Following the same logic to link ABM with regional water management

models (e.g., Colorado River Simulation System / WM-MOSART)

• Identifying the “upscaling” procedure for human decision-making

• Linking with POP and LULCC Thrust area

Thank you• Funding Acknowledgments: National Science Foundation Water

Sustainability and Climate grant #1360445. Department of Energy-Office of Science (Integrated Multi-Sector, Multi-Scale Modeling Scientific Focus Area). NREL Laboratory Directed Research and Development.

• Team:– NREL: Jordan Macknick, Matt O’Connell, Greg Brinkman– Sandia National Laboratories: Vince Tidwell– City University of New York: Charles Vorosmarty– Pacific Northwest National Laboratory: Nathalie Voisin, Tao Fu, Ian Kraucunas– USC: Measrainsey Meng, Kelly Sanders– Lehigh University: Ethan Yang – University of Colorado: Edith Zagona, David Neumann, Tim Magee– Los Alamos National Laboratory: Katrina Bennet

[email protected]/analysis/water

Questions?

61/11

Any Questions?

Develop a coupled

RiverWare-agent-

based model for the

San Juan River

Basin

62/11

Supplementary – ABM Structures

Internal Thinking Process – Bayesian Inference (BI) mapping

External Factors – Cost-Loss Ratio

𝑷𝒊,𝒕𝑩 = 𝒇 𝑷𝒊,𝒕−𝟏

𝑩 , 𝑷𝒕𝒇, 𝝀

• Past experience: water diversion, flow violation

• New incoming information: precipitation forecast

• Risk perception: 𝝀 ∈ [𝟎. 𝟓, 𝟏]

𝝀 = ቐ𝟎. 𝟓 ⇒ 𝑷𝒊,𝒕

𝑩 = 𝑷𝒊,𝒕−𝟏𝑩

𝟏 ⇒ 𝑷𝒊,𝒕𝑩 = 𝑷𝒕

𝒇

Risk-averse

Risk-seeking

𝒛 ≡expected cost of taking an action

opportunity loss if not taking an action

Agent’s (Farmer’s)

decision-making

𝑷𝒊,𝒕𝑩 ቊ

≥ 𝒛< 𝒛

expanding irrigation area

reducing irrigation area

63/11

Supplementary – Bayesian Inference Mapping

𝜞𝒑𝒓𝒕 =

𝝀𝚪𝒑𝒓𝒕−𝟏

𝝀𝚪𝒑𝒓𝒕−𝟏 + 𝟏− 𝝀 𝟏 − 𝚪𝒑𝒓

𝒕−𝟏𝚪𝒑𝒅𝒕 +

𝟏 − 𝝀 𝚪𝒑𝒓𝒕−𝟏

𝟏 − 𝝀 𝚪𝒑𝒓𝒕−𝟏 + 𝝀 𝟏 − 𝚪𝒑𝒓

𝒕−𝟏𝟏 − 𝚪𝒑𝒅

𝒕

If 𝝀 = 𝟎. 𝟓 𝜞𝒑𝒓𝒕 = 𝚪𝒑𝒓

𝒕−𝟏 decision is made fully relying on

past experience (belief)

If 𝝀 = 𝟏 𝜞𝒑𝒓𝒕 = 𝚪𝒑𝒅

𝒕 decision is made fully relying on

newly received information

64/11

Supplementary – Agent Groups

65/11

Supplementary – Agents’ Information

List of Model Parameters Preceding Factors of Agents

66/11

Supplementary – Extremity

Extremities Extremity

• reference of agents for

making decisions

𝑽𝒊𝒕 =

𝜽𝒊𝒕

𝜽𝒎𝒂𝒙− 𝟎. 𝟓

𝒊 ∈ 𝟏, 𝟐, 𝟑

Documents

Threats to Energy and Water Security for a Colorado River ... · 11/13/2018 · Case Study of a Provisioning Basin for Water and Energy: The San Juan Basin • A key source of water